1. Field of the Invention
The present invention generally concerns the diagnosis and treatment of tinnitus. The invention more specifically relates to the use of custom designed sound and feedback to determine the precise treatment sound matching the tinnitus, its neurophysiological effect, and monitoring the treatment effect by feedback from the brain.
2. Description of Related Art
Commonly perceived as a “ringing in the ear,” tinnitus is a very frequent disorder of the auditory system, affecting about 17% of the general population and up to 33% in the elderly. About a quarter of these people are sufficiently bothered by their tinnitus that they seek professional help [Jastreboff et al, 1996]. Tinnitus is a phantom perception and thus not associated with any auditory stimulus. Until very recently, there were no objective measurements that could be related to tinnitus [Jastreboff et al, 1994], and diagnosis of tinnitus had to rely on various questionnaires, e.g. [Wilson et al, 1991]. The fact that tinnitus is perceived as a sound, however, indicates that it is associated with aberrant neural activity in the auditory pathways. Furthermore, the fact that tinnitus is associated with perception leads to the conclusion that central auditory structures such as the thalamus and auditory cortex must be involved. Neural correlates of tinnitus have indeed been found in central auditory structures [Norena et al, 1 999; Mühlnickel et al, 1998; Wallhäusser-Franke et al, 1996]. Previously, tinnitus had been viewed as being caused in the auditory periphery [Eggermont, 1990; Tonndorf, 1981; Salvi and Ahroon, 1983], and even though neuronal activity related to tinnitus has been found in the central auditory system rather than in the periphery, it remains possible that the chain of events that leads to the development of tinnitus may be set off by events taking place in the periphery.
A Thalamic Model of Tinnitus
The presence of tinnitus-related neural activity in the central auditory system agrees well with a recently proposed neurophysiological model of tinnitus [Jeanmonod et al, 1996]. According to the model, tinnitus arises in the thalamus. Several parts of the thalamus interact to establish a reverberating loop. Neuronal activity originating from this reverberating loop is transmitted to the auditory cortex, where it gives rise to the tinnitus perception. This reverberating loop is established by disinhibition of neurons in the thalamus, which occurs when thalamic neurons receive inhibitory input causing a hyperpolarization. These thalamic neurons are capable of generating action potentials, once the hyperpolarization has reached a threshold. Thus, inhibitory input can cause these neurons to fire. The action potentials generated by this mechanism are called low threshold spikes (LTS) and usually occur in rhythmic bursts.
Two structures in the thalamus receive afferent auditory input: the medial geniculate body (MGB) and the multimodal medial thalamus (MT). Both these nuclei give excitatory input to the thalamic reticular nucleus (RT), and in turn receive inhibitory input from the RT. When afferent input into the MGB is normal, an equilibrium of excitation and inhibition is present in both MT and MGB. When the afferent auditory input into the thalamus fails, an imbalance of input is created. The MGB will receive no input whereas the MT receives a slightly decreased input. In this situation, the RT will still receive excitatory input from the MT, enough to generate the inhibitory output to both MT and MGB. Since no excitation is present in the MGB, the inhibitory input from the RT will hyperpolarize the neurons in the MGB. Neurons in the MGB will respond to this hyperpolarization with deinactivation leading to LTS bursts that are propagated to the auditory cortex.
Damage to the basilar membrane can set off the thalomocortical reverberating loop.
This model explains the appearance of tinnitus in conditions where the MGB is not receiving sensory input, e.g. in silent environments, or when hearing thresholds are temporarily or permanently raised after noise exposure. According to this model, once input into the MGB is restored, the tinnitus should vanish. Clearly, this is not the case in those people who are so annoyed by their tinnitus that they seek professional help. An additional mechanism must be in place to make tinnitus persist.
It has been noted that the presence of tinnitus cannot be determined from an audiogram, i.e. tinnitus may be present even when the MGB is receiving normal sensory input. The perception of tinnitus will only be stabilized when the person perceiving the phantom noise pays attention to it and associates it with unpleasant emotions [Jastreboff et al, 1996b; Langner and Wallhäusser-Franke, 1999] (FIG. 1). In this scenario, the person experiencing tinnitus is annoyed by it. Consequently, he will direct his attention to the phantom sound, thereby activating the limbic system. The limbic system is assumed to give rise to an increased generation of tinnitus-related activity. The detection of tinnitus-related activity is facilitated by mechanisms of lateral inhibition in the central auditory system. These will act to confine the phantom sound to regions representing distinct frequencies and increase the contrast between the tinnitus-related activity in these regions and the spontaneous activity in adjacent regions. This will lead to increased tinnitus perception. The increased perception can then lead to increased annoyance, completing a positive feedback cycle that will make tinnitus persistent.
The Auditory N100 As an Index of Cortical Responsiveness
Event-related potential (ERP) studies have demonstrated that electrical activity time-locked to stimulus or response events and averaged over repeated trials reflects information processing in the cortex [Hillyard et al, 1978; Pritchard, 1981b; Duncan-Johnson and Donchin, 1977; Pineda et al, 1997; Pineda et al, 1998]. Considerable evidence to date indicates that long-latency potentials (approximately >100 ms poststimulus) appear primarily sensitive to cognitive variables that reflect task requirements and the psychological state of the subject. These “endogenous” potentials most likely reflect non-obligatory activities invoked by the demands of the task.
One of the most widely studied endogenous components is the N100. Wolpaw and Penry [Wolpaw and Penry, 1975] were the first to propose that the N100 consisted of midline and temporal subcomponents. Subsequent work has shown that midline components can be modeled with tangential generators on the superior temporal plane and pointing toward the midline of the scalp. At least two midline components have been distinguished. An early frontocentral peak (N100a) shows reliable tonotopic changes in distribution and sensitivity to attention [Woods, 1995]. A later midline component (N100b) shows the same distribution for tones of different frequency.
N100 components show increases in amplitude and decreases in latency with increasing stimulus intensity [Hillyard et al, 1978; Scherg and von Cramon, 1990; Mäkelä and Hari, 1990]. In some individuals, the increases in stimulus intensity often bring decreased amplitude and increased latency. This tendency for N100 to increase or decrease in magnitude in response to stimuli of increasing intensity has been called the “augmenting/reducing” (AR) response. These intensity-amplitude functions have been hypothesized to result from variations in the central modulation of sensory processing and/or the actions of nonspecific arousal systems [Zuckermann et al, 1974; von Knorring and Perris, 1981]. It has also been proposed that they may reflect the “tuning” properties of a cortical gating mechanism that regulates sensory input [Buchsbaum and Silverman, 1968; Lukas and Siegel, 1977; Pritchard et al, 1985]. Some have related this mechanism to “attention” shifts and “overload” protection at high stimulus intensities.
A number of studies have shown that midline auditory N100 shows strong intensity dependence, while those recorded from temporal electrodes show weak intensity dependence [Pineda et al, 1991]. These differences suggest different N100 generators in the primary and secondary auditory areas [Pineda et al, 1991] [Connally, 1993], which is consistent with the multiple N100 generator hypothesis [Woods, 1995].
How Thalamo-Cortical Activity Affects the A/R-Response
The rhythmic LTS bursting activity of thalamic neurons mentioned above is also observed in slow wave sleep [Pape and McCormick, 1989; Steriade and Llinas, 1988], where it is thought to be a gating mechanism blocking sensory input into the cortex [Pape and McCormick, 1989; Steriade and Llinas, 1988]. The argument has also been made that this mechanism is active not only during sleep but also during waking and may result in different attentional states. The excitation of cortical tissue by tinnitus may compete with stimulus-induced activity. The neural activity induced by the tinnitus may, therefore, be regarded as competition for cortical neuronal substrates. This may lead to the reorganization of the auditory cortex [Mühlnickel et al, 1998]. That is, the processing of stimuli in the presence of tinnitus-related activity may lead to increases in the firing rate of neurons, the use of more neural substrate, or a combination of both. It is hypothesized that these mechanisms for dealing with tinnitus-related activity in the auditory system lead to the increased intensity dependence of the auditory evoked potential that Inventors and others have observed [Norena et al, 1999].
The hypothesis that tinnitus-related neural activity is caused by oscillatory LTS activity in the thalamus is further supported by a combination of other findings. First, application of serotonin in the lateral geniculate body (LGB) or MGB of the cat suppresses the hyperpolarization necessary to generate LTS-bursts [Pape and McCormick, 1989]. Second, the intensity dependence of the auditory evoked potential is strongly influenced by brain serotonergic activity [Juckel et al, 1999; Juckel et al, 1997; Hegerl et al, 1996]. Finally, these observations are linked by the fact that the thalamus contains a high density of binding sites for serotonergic drugs as well as serotonin uptake sites [Smith, 1999].
Elaboration on the Tinnitus Models
Taken together the various models and evidence suggest that while tinnitus may be triggered by events in the periphery, the mechanisms that make tinnitus a persistent and annoying condition are located in the central auditory system. Furthermore, it appears that people suffering from tinnitus unintentionally train themselves to have tinnitus by using negative reinforcement. It has been shown that the cortical representation of tones associated with unpleasant sensations is enlarged [Gonzalez-Lima and Scheich, 1986] and/or changed to enhance the contrast between this particular tone and other tones of similar frequency [Ohl and Scheich, 1996]. The cortical representation of tones no longer associated with unpleasant sensations will return to a state where contrasts are no longer enlarged. Moreover, responses to stimuli occurring while attention is directed at another task will decrease over time [Anderson and Oatman, 1980]. This suggests that if the association between tinnitus and unpleasant emotions can be broken, the aberrant neuronal activity in the central auditory system can be decreased by habituation. Then tinnitus is treated like any other sound that does not carry relevant information: it is ignored. The tinnitus retraining therapy (TRT) introduced by Jastreboff [Jastreboff et al, 1996a] makes use of the mechanism described above. However, TRT as described by Jastreboff uses white or broadband noise as a habituating stimulus. The rationale behind the use of white noise is to generate a decreased signal-to-noise ratio between the tinnitus-related neuronal activity and random background activity in the auditory system. This would be achieved by introducing a quasi-random, stimulus-driven activity into all of the parallel tonotopic channels of the auditory system.
A precise computational model of tinnitus has been proposed by Langner et al [Langner and Wallhäusser-Franke, 1999] based on animal work. This model assumes that the limbic system is necessary for stabilizing the tinnitus perception. It also explains how a decreased auditory input resulting from a peripheral hearing deficit can give rise to a specific tinnitus pitch. When the tinnitus sound is used as a habituation stimulus, Langners' model predicts the tinnitus would disappear.
Currently, there are sound therapies for tinnitus that use generic sounds. These present therapies are only partially effective and require a long time for treatment. It would, therefore, be desirable to obtain a brain signal feedback system, wherein one could rapidly suppress brain activity related to tinnitus and provide relief for this disorder.